Why the doubts over the Rotavac anti-diarrhoea vaccine may be overblown

By Priyanka Pulla

A strange controversy has been brewing over the world’s cheapest rotavirus vaccine, Rotavac, developed by Indian researchers and included in the country’s Universal Immunisation Programme this month. The vaccine was an important breakthrough for India: rotavirus, a wheel shaped pathogen (hence its name) that spreads through food and drink contaminated with infected faeces, causes severe diarrhoea among children. In India, it kills nearly 80,000 children annually, and leads to a million hospitalisations. And yet, India hasn’t had an affordable vaccine so far. Existing rotavirus vaccines from Glaxo Smithkline and Merck cost Rs 2,398 and Rs 2,700 a course, placing them squarely out of reach of many Indians.

Representational image. AFP

Representational image. AFP

So, it was a cause for celebration when the results of a clinical trial for establishing the effectiveness of the vaccine were published in the medical journal Lancet in March 2014. Ever since this happened, however, a few researchers, prominent among them Jacob Puliyel, a pediatrician from Delhi’s St Stephen’s Hospital and a member of the National Technical Advisory Group on Immunisation (NTAGI), have expressed concerns about the safety of the vaccine.

In a letter to the editor of the journal Vaccine, Puliyel has argued that the trial data points to a high risk of intussusception---a potentially fatal side-effect of rotavirus vaccines, in which the intestine folds into itself, causing abdominal cramps, bloody stools, nausea and vomiting. Further, he is asking for unpublished data on the trial, which, he says, will reveal if intussusceptions happened more frequently at the Vellore centre of the vaccine trial, compared to the Delhi and Pune centres. The demand was echoed in a petition posted by several others on 12 May on the campaign website Avaaz.com.

But the research team that carried out the Rotavac trial, as well the Department of Biotechnology (DBT) that sponsored it, are categorically saying that Puliyel’s arguments are wrong. And since they are wrong, DBT secretary K Vijayaraghavan says, the DBT doesn’t feel the urgency in disclosing the unpublished data. “We do plan to discuss the issue and similar data requests, but that will happen when time permits. In the absence of concerns by any recognised safety experts, I see no urgency and would prefer a considered approach after appropriate consultations,” he wrote in an email to this journalist.

Which side is one to believe?

What the Rotavac trial really found

To understand the issue, readers need to know a little more about how the clinical trial for Rotavac was conducted and what it found. The trials for Rotavac began in 2011, when the researchers identified 6,799 babies of about six weeks of age for participation. Out of them, 4,532 received three doses of the vaccine till about three months of age, while 2,267 received three doses of a placebo, or saline injections. The rate of gastroenteritis in the placebo and vaccine groups was then compared.

After the third dose of the vaccine was given, the researchers continued to monitor the children very closely for intussusception until they were around two years old, an unusual practice for rotavirus vaccine trials. The trials for Merck’s Rotateq, for example, had looked for intussusception for only a 42-day period after each dose of the vaccine, and thereafter, every six weeks for a year after the first dose.

The Indian researchers were also far more careful in monitoring and treating children for intussusception than their international counterparts were - something to be applauded, given the recent controversies surrounding clinical trials in India. Health workers in the Rotavac trial contacted the parents of a third of the children enrolled in the trial every single day for 14 days after all three doses of the vaccine. The remaining parents were contacted weekly till their children were two years of age. The parents were also given cell phones and round-the-clock access to the researchers so that any troubling sign of intussusception could immediately be tackled. Symptoms such as vomiting or a distended stomach, which don’t always mean intussusception, were further investigated with ultrasound imaging.

The researchers kept up this surveillance for two years, far beyond the period of 14 days in which vaccine-related intussusceptions are known to occur. “This has never been done before,” says Gagandeep Kang, professor of gastro-intestinal sciences at Christian Medical College, Vellore, who was part of the study. “We were dragging kids out of their houses. Parents were fed up of this. If the child had blood in the stool, we were taking them to the hospital to do ultrasound investigation.” In short, it appears that extreme care was taken in monitoring children compared to other vaccine trials. This is key to understanding what the Rotavac controversy is about.

After the two year period, Kang and team published their study results: children who received the placebo had fallen ill with rotavirus gastroenteritis 53.6 percent more often than those given the vaccine. Clearly, the vaccine was working much better than the placebo. But the researchers also noted that among the 4,532 vaccinated children, 17 developed intussusception, while 6 did so in the placebo arm. These rates of intussusception were detected using ultrasound after the child showed any of a broad list of symptoms mentioned above.

The problem with doing an ultrasound on children having broad symptoms such as vomiting is that it picks up several minor or transient cases of intussusception that do not need treatment and go away on their own. (It’s a bit, although not entirely, like giving a tuberculosis blood test to everyone with a cough. One third of the world’s population has latent tuberculosis, which shows up in blood tests, but doesn’t produce symptoms or spread to others. Only 10 percent of these go on to develop actual tuberculosis.)

The real rate of intussusception, therefore, was calculated by Kang’s team using a much more sensitive set of diagnosis criteria, called Brighton Level 1. These were labelled “confirmed intussusceptions”, and Kang’s team found only eight in the vaccine arm, compared to the three in the placebo arm, a statistically insignificant difference. So, they concluded, the vaccine wasn’t causing more intussusceptions than the saline solution.

That wasn’t the only logic they relied on to exonerate the vaccine, though. It so happens that intussusception, with no obvious cause, occurs every once in a while among children of the age enrolled in the trial. In parts of the world, such as Vietnam, one out of every 300 children develops intussusception, and this isn’t due to any vaccine - it is simply an ailment that the very young are susceptible to. The baseline rate of intussusception isn’t known for India, because we haven’t carried out any large scale studies yet, but intussusception occurs among children here too. Therefore, the only way one can tell if intussusception was caused by a vaccine is if it occurs within one, two or a maximum of three weeks after a dose.

Readers should now note the second key fact of the Rotavac controversy: None of the children given the vaccine in the Rotavac trial developed intussusception within a month of the first three doses. In fact, the first child to develop intussusception did so in the placebo group, 36 days after the third dose of the saline injection. The first vaccine recipient to develop intussusception got it 112 days after the first dose. Hence, the researchers concluded, and with pretty good reason, that there was no link between the vaccine and the intussusception.

In addition to total intussusception rates across the trial sample, the researchers also published a breakdown of intussusception rates by trial centres - 581 per 100,000 child-years at Vellore, 178 per 100,000 child-years at Pune and 27.7 per 100,000 child-years at Delhi. These numbers are extrapolations based on the actual number of intussusceptions seen over two years for the 1,500 children each enrolled in Vellore and Pune and the 3,799 children enrolled in Delhi.

Why this data looks troublesome (but really isn’t)

So what’s the controversy about?

Puliyel and the other signatories of the Avaaz petition have made two major arguments. First, they say, the numbers 17 and 6 point to a high rate of intussusception compared to data from older vaccines. Second, Puliyel has argued, the intussusception rate of the Vellore arm, at 581/100,000 child years, is huge compared to the Pune (178/100,000 child years) and Delhi (27.7/10,000 child years) centres.

So, the petition demands that Christian Medical College, which coordinated the vaccine trial, release the number of intussusceptions in the placebo group and the vaccine group of the Vellore centre, so that the difference in both arms can be compared and tested for statistical significance. The fact that the DBT has not released this bit of data despite repeated requests by Puliyel makes him even more wary. “There has been a very severe reluctance to provide this data,” he said in an interview with this journalist. “This makes one suspicious that, probably, there was a statistically significant difference even in that very small sample. If there is a statistically significant difference (between the vaccine and placebo arms) in that small sample (of Vellore), obviously the vaccine cannot be used, because you do not know which other populations are similarly susceptible.”

Let’s look at both arguments separately. The first argument, that 17 out of 4,799 is a high number of intussusceptions is based on a comparison with data from Wyeth’s Rotashield vaccine. In 1999, about a year after it was launched and administered to around 1.5 million children, Rotashield was withdrawn from the market when post-launch surveillance showed that was causing 1-2 excess intussusceptions for every 10,000 vaccinated children, compared to the unvaccinated. This happened even though pre-launch studies among 14,687 children threw up no red flag for side-effects.

Puliyel and other signatories of the Avaaz petition have compared the 11 excess intussusceptions in the vaccine arm of the Rotavac trial (17 minus 6) with the 1-2 intussusceptions identified from Rotashield’s post-launch data of over a million children, saying Rotavac seems to be causing 5-10 times more intussusceptions than the withdrawn Rotashield did. “That is slightly alarming,” Puliyel says.

But such an extrapolation has no meaning if one remembers the difference in samples of both trials. Rotavac was given to a mere 4,532 children, whereas, by the time Rotashield was withdrawn, it was given to more than a million. The reason the trial for Rotavac is too small to detect intussusceptions is because it occurs only once in over 20,000 cases (as we know from the experience of Rotashield, Rotateq and Glaxo Smithkline’s Rotarix). In statistical terms, such a trial is said to be ‘underpowered’ to detect a rare adverse event such as intussusception. That’s why, even a 14,678-infant trial for Rotashield threw up nothing, leading to its launch.

The second issue brought up by Puliyel relies on a similar extrapolation of the numbers from the Vellore, Pune and Delhi arms. He is right when he says that the 587/10,000 intussusception rate of the Vellore arm is more than 20 times that of the Delhi arm. This could mean one of two things: one, the baseline rate of intussusception in Vellore is naturally extremely high (as is the case with Vietnam). Or, two, the vaccine is, for some regional factor, causing very high rates of intussusception in Vellore.

The problem here is: if 6,799 is a small sample size to detect intussusceptions, the 1,500 infants in the Vellore trial is even smaller. Also, when a clinical trial is designed to analyse aggregate data from several different trial centres, analysing data from each centre separately can distort the picture. According to Sylvie Escolano, a bio-statistician at France’s INSERM who wasn’t part of the Rotavac study, “From a statistical point of view, the trial was designed to be a metacentric trial; only the results issued from the analysis of the whole data are valid.” The WHO seems to agree, as it, too, has declared Rotavac safe based on current data.

Even if there is a statistically significant difference in the intussusception rates between the placebo arm and the vaccine arm in Vellore, this data cannot be used to say the side-effect occurs more often in Vellore. A trial to show this will have to be designed differently and will have to be larger.

If trial isn’t large enough to address safety, why is the vaccine being launched?

Testing if a vaccine is causing intussusception is a big problem the world over, because intussusception is such a rare event. Muddying the waters further is the fact that it happens naturally among children, making it hard to implicate the vaccine. The only way to be absolutely sure that a vaccine is causing intussusception is to do a massive trial of over 100,000 children. ”It is estimated that the approximate size of a study that actually estimates rate of intussusception is close to half a million,” says Kang. This is why, neither the makers of Rotashield, nor Rotateq, nor Rotarix were able to truly assess intussusception at the pre-launch phase, even though the Rotateq and Rotarix trials enrolled as many as 60,000 children each.

Why, then, were they allowed to launch the potentially dangerous vaccines? The answer lies in the risk-benefit ratio. The WHO calculates that because an effective vaccine could save the lives of thousands in a country such as India, an extremely rare event such an intussusception should not be a reason to withhold launch. Instead, vaccine developers are allowed to launch vaccines after small trials, as long as they put in place strong surveillance mechanisms that can track intussusceptions, and ensure the affected children are treated. It is precisely such a vaccine adverse-effect surveillance mechanism that helped identify the intussusception rates of Rotarix, Rotateq and Rotashield, and caused Rotashield to be withdrawn. Meanwhile, Rotarix and Rotateq continue to remain part of national immunisation programmes of 77 countries, because their risk-benefit ratio strongly favours this, even though they too cause low levels of intussusceptions.

Should you worry about Rotavac?

Against this background, the launch of Rotavac in India and its inclusion is not problematic at all, as long as the health ministry ensures that intussusceptions are tracked and treated. And the fact that none of the intussusceptions in the pre-launch trial happened in the window after the vaccine was administered is cause for reassurance. “The vaccine in this small number is completely safe,” says Thekkekara Jacob John, former professor of virology at Christian Medical College, Vellore, who wasn’t part of the Rotavac study. In fact, John found the lack of vaccine-associated intussusceptions in this trial so remarkable that he has come up with three hypotheses for why this is the case. One of them is that the Rotavac vaccine strain is a natural neonatal strain which is not pathogenic to young children, compared to the strains used in previous vaccines.

Yet, given past experience with rotavirus, it cannot be said for sure that Rotavac doesn’t cause intussusception. Only larger trials will tell. This is why the DBT has planned post-launch trials of one lakh children, in which they will look for a rate of intussusception of one in 20,000 children. Kang warns that even such a large trial may not be enough. “We are looking for an intussusception range of one in 20,000. If the rate is higher, we will definitely pick it up. If it is lesser, we may not be able to pick it up even with this size.”

In the meantime, the Universal Immunisation Programme, which has included the vaccine, will need to watch children closely for intussusception. Puliyel is right about one thing - while mortality rates for intussusception in the west are only about 1 percent, the situation in India is different because of our pathetic public healthcare infrastructure. “The mortality in rural areas is about 30 percent,” he explains. There is no doubt that the health care givers providing the vaccine will need to be cautious. This is the real challenge.

Should DBT disclose the data?

World over, there is a move towards transparent clinical trials, with the British Medical Journal (BMJ) making it mandatory for researchers to disclose data upon reasonable request, and the WHO supporting the registration of all trials at initiation, so that negative results aren’t suppressed. So, why shouldn’t the DBT disclose the data, given that it is a government body that funded the trials for a vaccine that is now a part of the Universal Immunisation Programme? This is definitely a good idea, except for the fact that it isn’t clear what the new data will show. Even the BMJ suggests that a researcher requesting raw data should submit a protocol for re-analysis of the data (that will, presumably, alter the results). Perhaps Puliyel and other signatories of the petition should explain how exactly the new data will alter our understanding of the vaccine’s effects.


Published Date: Jul 18, 2015 07:36 pm | Updated Date: Jul 20, 2015 11:31 am



Also See